Detail

Fine-grained parallelism on modern computing architectures

Parallel computing architectures have undergone several deep transformations over the last ten years or so, with the spreading of multi-processors followed by multi-cores processors and now followed by the many-cores processors, the introduction of non-uniformity in processor-memory accesses, and more recently the introduction of non-uniformity among the processors and cores themselves as with the IBM Cell or the general purpose GPUs. This talk will present some background, as well as current and future works of the INRIA Runtime team on providing efficient, portable runtime systems and software environments to help programming such architectures and keep-up with the evolution.